Study of anomaly detection techniques in exchange rate time series data
Νικόλαος Ευθυμίου Μπάμπης · Aristotle University of Thessaloniki · 2020
Time series outlier detection has been attracting a lot of attention in research and application. In this diploma thesis, we introduce the problem of detecting outliers in exchange rate data set based on distance and density methods. We study the problem using Euclidean distance in distance-based algorithm and Local Outlier Factor (LOF) for density-based approach. The problem is analyzed with big data terms. So, the usage of Apache Spark Framework for the implementation was a key parameter of the thesis. The overall aim of the diploma thesis is to examine anomaly detection in time series with big data conditions.